David Phillippo

R for Health Technology Assessment R for Health Technology Assessment discusses the use of proper statistical software, specifically R, to perform the whole pipeline of analytic modelling in health technology assessment (HTA). It has b...

Hard copies and eBook version available from Chapman & Hall / CRC

www.routledge.com/R-for-Health...

R for Health Technology Assessment - new book out today!

Free online version gianluca.statistica.it/books/online...

The marginal_effects() function wraps predict() to create differences or ratios of absolute predictions.
For example:
* risk differences/ratios from an analysis of log odds ratios
* marginal differences in RMST or time-varying marginal hazard ratios from a survival analysis

📣 multinma update v0.7.1 on CRAN * New marginal_effects() function for computing marginal relative effects
* Progress bars for long operations
* trt_ref argument to predict() has been renamed to baseline_ref for consistency
* Bug fixes Full details 👉https://t.co/abYabQCvKS

This is now resolved: Stan has been patched, and multinma is back on CRAN
https://x.com/dmphillippo/status/1765397920130510965?s=20

multinma is back on CRAN 🎉

Stan has been patched to fix the memory allocation bug

This release (v0.6.1) also includes a bugfix for piecewise exponential hazards models - changelog here 👉
https://t.co/abYabQCvKS

Binaries will be built by CRAN over the next few days https://t.co/4UHJPzfgig

multinma will be back on CRAN as soon as rstan is patched - or sooner if CRAN respond to my emails requesting reinstatement!

More details on the memory allocation bug here 👉 https://github.com/stan-dev/rstan/issues/1111

If you need to install, you can use R-universe:
install.packages("multinma", repos = c("https://dmphillippo.r-universe.dev", getOption("repos")))

Automatic checking of integration error for ML-NMR:
- Checks sufficient number of integration points within a single model run
- Gives nice warnings if action is required
- Much lower default n_int=64 (previously 1000!) means much faster models!

Multilevel network meta-regression for general likelihoods: synthesis of individual and aggregate data with applications to survival analysis Network meta-analysis combines aggregate data (AgD) from multiple randomised controlled trials, assuming that any effect modifiers are balanced across populations. Individual patient data (IPD) meta-regression is the "gold standard" method to relax this assumption, however IPD are frequently only available in a subset of studies. Multilevel network meta-regression (ML-NMR) extends IPD meta-regression to incorporate AgD studies whilst avoiding aggregation bias, but currently requires the aggregate-level likelihood to have a known closed form. Notably, this prevents application to time-to-event outcomes. We extend ML-NMR to individual-level likelihoods of any form, by integrating the individual-level likelihood function over the AgD covariate distributions to obtain the respective marginal likelihood contributions. We illustrate with two examples of time-to-event outcomes, showing the performance of ML-NMR in a simulated comparison with little loss of precision from a full IPD analysis, and demonstrating flexible modelling of baseline hazards using cubic M-splines with synthetic data on newly diagnosed multiple myeloma. ML-NMR is a general method for synthesising individual and aggregate level data in networks of all sizes. Extension to general likelihoods, including for survival outcomes, greatly increases the applicability of the method. R and Stan code is provided, and the methods are implemented in the multinma R package.

Accompanied by our latest preprint: https://arxiv.org/abs/2401.12640

https://x.com/dmphillippo/status/1750486767570981227?s=20

More survival analysis:
- Left/right/interval censoring, delayed entry
- Predict and plot survival probabilities, hazards, cumulative hazards, mean survival times, restricted mean survival times, quantiles of the survival time distribution, and median survival times

📣multinma v0.6.0 update on CRAN

Major new features (details below):
- Survival analysis
- Automatic integration convergence checking (faster models!)

Plus other improvements and bugfixes

Full changelog 👉https://dmphillippo.github.io/multinma/news/

More nuggets in this paper:
- A new algorithm for automatic convergence checking for numerical integration 👉 fewer integration samples needed, nice warnings, MUCH faster ML-NMR models
- M-spline baseline hazard model with a novel random walk shrinkage prior 👀

Survival analysis with multilevel network meta-regression? Yes please! New preprint extending ML-NMR to likelihoods of any form, including for survival analysis. Accompanied by a new multinma release v0.6.0, which is on CRAN now. https://arxiv.org/abs/2401.12640

📣 multinma update 0.4.2 is on CRAN

Fixes a couple of bugs when trials have repeated arms of the same treatment 🐙
✅ get_nodesplits() for node-splitting no longer errors
✅ printing the network now shows the repeated arms

Details 👉 https://dmphillippo.github.io/multinma/news/index.html

📣 Bugfix update multinma 0.4.1 rolling out on CRAN

Fixes an issue introduced with tidyr 1.2.0 that broke ordered multinomial models

Details 👉 https://dmphillippo.github.io/multinma/news

📣 Update to multinma v0.4.0 on CRAN
- Node-splitting for checking inconsistency
- Predictive distributions for random effects models
- Improved handling of correlations for integration points (ML-NMR models)
- And more! Details 👉 https://dmphillippo.github.io/multinma/news
#rstats #metaanalysis

Booking now open for our network meta-analysis course 👇 https://x.com/sdias_stats/status/1486706844466827267

PhD opportunity in Glasgow - still time to apply!

Predictors of early trial termination using individual-level participant data and aggregate-level data from multiple trials

Co-supervised by myself, advisory team includes @sdias_stats and @WeltonNicky

https://t.co/h4USGbeADd

📣Update to multinma (v0.3.0) now on CRAN

- New features for flexibly specifying baseline distributions when producing absolute predictions
- Squashes bugs when specifying certain types of models with contrast data

Full details: https://dmphillippo.github.io/multinma/news/
#rstats #metaanalysis

Looking forward to speaking at @HERC_Oxford this Wednesday - details and registration at the link below https://x.com/HERC_Oxford/status/1373967203851251715

multinma An R package for Bayesian network meta-analysis of individual and aggregate data. Presented at ESMARConf 2021.

Slides from this talk are now online too: https://dmphillippo.github.io/ESMARConf2021_multinma/

ESMARConf2021 livestream Session 2B - Quantitative Synthesis (NMA)

Enjoyed presenting the {multinma} package at #ESMARConf2021 yesterday - if you missed it the recording is available on YouTube: https://youtu.be/d4ufa__hGbY?t=652

#metaanalysis #rstats https://x.com/eshackathon/status/1352291884941664259

Catching up on @cantabile's excellent #ESMARConf2021 talk from earlier this morning, developing NMA reporting toolchains for stakeholders like Cochrane. Great to see {multinma} and {nmathresh} being used in the wild too! https://x.com/eshackathon/status/1352529752419164160

📣Update to multinma (v0.2.1) now on CRAN

- Squashed a couple of bugs
- Improved documentation of available likelihoods and link functions

Details: https://dmphillippo.github.io/multinma/news/

📣 Update to multinma (v0.2.0) released

Changes include:
- Models for ordered categorical data + example vignette
- Overview of examples for easier navigation
- Inline data transformations
- Improved efficiency when working with fitted models

Details: https://t.co/abYabQCvKS

The {multinma} R package now has a website!
👉 https://dmphillippo.github.io/multinma/ 👈 - All documentation with illustrated code
- Walkthroughs of example analyses #rstats #metaanalysis

Methods covered include: network meta-analysis, population adjustment, combining observational and randomised evidence, multiple outcomes, surrogate outcomes, survival outcomes, reliability of recommendations, and comparative efficacy of diagnostic tests https://t.co/QqzaU1ilBp

📣 New paper! #openaccess Comparing the performance of population adjustment methods (MAIC, STC, and multilevel network meta-regression) in an extensive simulation study https://onlinelibrary.wiley.com/doi/10.1002/sim.8759

Very spoilt by my lovely wife with this birthday gift! 🎁 Stunning print from @thomasp85, thank you!